BicGO: a new biclustering algorithm based on global optimization

Abstract

Recognizing complicated biclusters submerged in large scale datasets (matrix) has been being a highly challenging problem. We introduce a biclustering algorithm BicGO consisting of two separate strategies which can be selectively used by users. The BicGO which was developed based on global optimization can be implemented by iteratively answering if a real number belongs to a given interval. Tested on various simulated datasets in which most complicated and most general trend-preserved biclusters were submerged, BicGO almost always extracted all the actual bicluters with accuracy close to 100%, while on real datasets, it also achieved an incredible superiority over all the salient tools compared in this article. As far as we know, the BicGO is the first tool capable of identifying any complicated (e.g., constant, shift, scale, shiftscale, order-preserved, trend-preserved, etc), any shapes (narrow or broad) of biclusters with overlaps allowed. In addition, it is also highly parsimonious in the usage of computing resources. The BicGO is available at https://www.dropbox.com/s/hsj3j96rekoks5n/BicGO.zip?dl=0 for free download.

Indexing and Search of Order-Preserving Submatrix for Gene Expression Data

“…Wang et al [31] apply the longest common subsequence (LCS) framework to selected pairs of rows in an index matrix derived from an input data matrix to locate a seed for each bicluster to be identified. Li and Su [32] introduce a biclustering algorithm called BicGO to recognize complicated biclusters submerged in large scale datasets (matrix). Jiang et al [33] proposes two constrained OPSM query methods, which exploit user defined constraints (must-link, cannot-link, interval, count) to search relevant OPSMs from two kinds of indexes introduced.…”